How clinicians use data to make an impact on clinical outcomes Andrew Brodbelt Consultant Neurosurgeon and Clinical Director of Neurosurgery, The Walton Centre NHS Foundation Trust Member of CNS SSCRG, National Cancer Intelligence Network

The core objective: link data with patient outcome Promoting data collection National repository datasets Expert analyses Improve standards of care and outcomes Support audit and research

Key issues & priorities Ownership: Develop, encourage, QA Analytical programme Reporting Communication Research Clinical engagement

Audit Peer Review Research Revalidation What clinicians use data for

My Journey Neuro-oncology lead MDT Chair CNG Chair Interested in Incidence and outcome – Regional data NCIN

What is the outcome in the UK for patients with glioblastoma? Commonest primary malignant brain tumour Malignant incidence 5.3/100,000 West Median month

UK national data rare. Jan 2007 – Dec 2011 National cancer registration service Hospital Episode Statistics (HES) NHS Personal Demographic Service Radiotherapy data not complete until 2012 England only

Age adjusted incidence: 4.64/100,000

Overall survival remains low 6.08 months 28.4% 3.4%

Patient age is important 16.2 months 7.9 months 3.2 months

The effect of treatment

Maximal Treatment improves survival at all ages European Journal of Cancer, 2015

How can this help with my clinical management?

Need to increase the numbers treated aggressively Over 70 years – Maximal treatment if possible – If not discuss no treatment Is anyone cured?

Some patients survive long term

17,786 patients with Glioblastoma Deduplication 938 Case by case 484 Possible 256 Definite 96

Pathology 94 Glioblastoma –? GBM-O 2 Giant cell glioblastoma

Young age, and equal sex

Treatment profile 70% VS 25%

48 % had multiple operations. DebulkingsPatient Numbers

A late recurrence rate 5 year survival 3.4% from diagnosis, but 36% at 5 years

Most patients (2/3rds) who survive 5 years will still die of their disease within 10 years of diagnosis

There are study limitations True Glioblastoma patients? –Need independent check of pathology Earlier years data sparce Radiotherapy data assumed Not the complete cohort –What about the other 256? Quality of Life

What does this tell us? Aggressive treatment can lead to longer outcomes? There are some data limitations The next step Is prolonged survival due to the patient, the tumour, the treating clinicians or a combination? Registered as a national audit

30 day mortality and the ‘50% rule’

30 day mortality after surgical resection of a brain tumour Dr. Matt Williams ICHNT & IC On behalf of the NCIN Brain & CNS SSCRG

30-day post-operative mortality against # operations in 3 yrs...

... plotted on logged horizontal scale...

... and jittered to show individual surgeons

Enrichment very skew distribution of Load: – 152 surgeons with only one operation – 73 with from one to five operations – 0 with six operations – 228 with seven or more operations interest currently in surgeons who operate regularly, so dataset enriched by excluding – surgeons with fewer than six operations – surgeons without at least one observation in both the first and last six months of the three year period

Results Originally patients and 453 surgeons at 31 trusts – 152 surgeons only did one operation, and others did a few 9194 (84%) patients and 163 surgeons (36%) in 30 trusts Predominantly brain & meninges – Other rare 30 day mortality was 3%

Results Lowest surgical activity = 7 Median number = 46 (over 3 years) Quartiles: – 7 – 29 – 29 – 46 – 46 – 70 –

Original dataset...

... enriched dataset...

... re-scaled...

... and most of jittering removed

... and prediction from logistic model

Results Age, deprivation and individual surgeon volume correlated with 30 day mortality – Patient sex & trust volume were not Same factors preserved on multivariate, on both step-forward and step-backwards analysis of factors RR for surgeon volume is 0.8, p = – 20% relative risk reduction in 30 day death for doubling a surgeon’s workload

Conclusions There is a surgical volume effect There is probably not a centre volume effect – Because we are already centralised Moving patients from the least busy ¼ of surgeons to the others (from LVS to MVS): – Needs proper modelling – ~700 pts who had surgery with a LVS – Moving them to a MVS turns their absolute risk from ~3% - > 2.4% 4 lives 30 days THESE CALCULATIONS ARE VERY PROVISIONAL

Where to from here? Meningiomas Long term survivors Details of radio and chemotherapy Papers, presentations, and national committees to drive improved care

Get Involved Improve data collection Ask the important questions Use the data to drive clinical change. At least one clinician per MDT.